SP VL Paluck

1.      An uptick in prejudice reduction research (some of which will become classic)

2.      The modal research is very different from these future classics  

3.      There are many reasons not to trust the majority of the research

-> some of the most popular prejudice reduction ideas are not supported  4. The most rigorous research shows very small reduction in prejudice  

➔  what should the next generation of prejudice reduction research look like?  

➔  Are we using the right model of change?  

Prejudice Reduction Research from 2007 to

2019

▪  General uptick showed via the black line referred to as “total”  

▪  Uptick mostly for lab and online studies

▪  ups and downs for field studies, mostly decreasing   

Biomedical meta-analytic standards 

▪  to show that datas and research is actually replicable  

▪  5 searched separate databases  

▪  over 16.000 results narrowed to 1835 reviewed in full  

▪  Criteria: prejudice as animus (- sexism, partisan & collegiate bias), experimental  

▪  Final sample: 418 experiments in 309 manuscripts from 2007 – 2019  

▪  All experiments coded for quant findings and qualitative features of study by 2 coders  

▪  Robust, realistic interventions involving social as well as psychological processes

▪  Robust methods to test them: large sample sizes, measure behavior as well as attitudes, attention to randomization, pre-registration, open data  

▪  Despite varied theoretical approaches, all studies have promising if limited / small positive outcomes  

-> e.g. Team sports as intergroup contact: Lowe (2020) & Mousa (2020)

-> e.g. Confronting people who use racial slurs (on twitter): Munger (2017)  -> e.g. Does online diversity training work? Chang et al. (2019)

Prejudice reduction through mentalizing

-> individualized approach  

Light touch interventions (76%)  

-> coding category

-> easy to implement  

Methodological problems: small sample sizes, attrition, cluster randomized, lack of transparency  

▪  d = .357 (Cohen’s d is an effect size used to indicate the standardized difference between two means)  

▪  SD = .02  

➔  very small sample sizes!  

▪  shows the same quantiles as before but in a graphical way  

▪  more people = lower effect sizes 

▪  average effect per category of every intervention  

-> for the entire sample    

▪  only looking at the studies with top quintile in terms of sample soze for each category  

▪  affects every category of research    

➔  Diversity training  

➔  Implicit bias: training, prediction of attitudes & behavior  

➔  Small changes building over time  

▪  prediction line crosses 0  

▪  based on used population data  - use all best methods (large sample sizes etc.)

- suggests that if we follow the same average approach that we’ve been taking, the most rigorous studies would find no change in prejudice

➔  Many recommendations in the paper (annual review 2021) for those interested in field or laboratory work  

➔  a bolder step forward (and less developed): they invite the field to think about a different model of change, and more about intervening in structures  

Current model of change:

➔  Attack a psychological problem with individualized psychological interventions, to create individual psychological and societal change  

Alternative model of change:

➔  Attack a psychological problem with structural interventions, to create individual psychological and societal change  

▪  Structure: institutions, rules, leaders  

▪  Changing laws, rules, or organizations; decisions and communications from leaders  

BUT: also think about social structures

▪  mass collective experiences, unofficial organizations; mass media events, large in person gatherings, simultaneous collective experiences  

▪  Structurally-related theory: social norms theory predicts leaders can signal new social norms about the unacceptability of prejudice

-> structural interventions: supreme court decisions (Tankard & Paluck, 2017; Clark, Martinez & Paluck, in prep)

-> less structural intervention: send emails to people reminding them of progressive orientation of their leader

▪  Individually-oriented theory: perspective taking theory predicts perspectives of outgroup members can reduce prejudice

-> structural intervention: Brookman & Kalla canvassing intervention  

-> less structural intervention: test nudges to stimulate perspective getting  

▪  Other structural intervention examples  

▪  Improve our skills at thinking about structural expressions of our theories, designing structural interventions  

▪  Collaborations: a more truly interdisciplinary social science

▪  Research to understand how these interventions could move forward (backlash research?)  

▪  Willingness to push for equity reforms as interventions

▪  Use a variety of causal identification strategies like regression discontinuity, instrumental variables, other natural experiments